For twenty years, the funnel has been the default mental model of growth: pour prospects in at the top, convert a fraction at each stage, collect customers at the bottom. It is intuitive, it maps neatly to org charts, and it is quietly holding a lot of startups back.

The alternative model — the growth loop — reframes growth as a closed system where the output of one cycle becomes the input of the next. The difference sounds academic. It is not. It is the difference between linear growth you must keep paying for and compounding growth that pays for itself.

The structural problem with funnels

A funnel is an open system. Users enter at the top, and value exits at the bottom. Three consequences follow:

Funnels consume; they do not produce. Every new customer requires new input — more ad spend, more content, more outbound. Stop feeding the top, and growth stops within a sales cycle.
Funnels degrade. Channels saturate. Ad costs rise as you exhaust your best audiences. The same funnel that acquired customers at 50 units of cost last year acquires them at 80 this year.
Funnels fragment ownership. Marketing owns the top, product owns the middle, sales owns the bottom. Nobody owns the system, so improvements are local and often conflicting.

None of this means funnels are useless. Funnel math is still the right way to measure conversion between two points. The problem is using the funnel as your growth strategy.

What a growth loop actually is

A growth loop is a closed sequence of steps where users or content generated in one pass feed the next pass. The canonical shape:

A new user joins.
They take an action inside the product.
That action produces an output visible to non-users (an invitation, a shared artifact, a public page, a data contribution).
The output brings in new users, and the cycle repeats.

Classic illustrations: a document tool where every shared doc exposes recipients to the product; a scheduling tool where every booking page is a mini landing page; a marketplace where every seller recruits buyers and every buyer attracts sellers.

The critical property is compounding. If 100 users generate outputs that bring in 20 new users, those 20 bring in 4, then 1 — one cohort yields roughly 125 users total from 100 acquired. And unlike a funnel, the machinery keeps running without new budget.

Loops versus funnels: the math

Consider two startups, each starting with 1,000 users a month from paid channels (illustrative numbers):

Funnel company: every month it buys 1,000 users. After 12 months: 12,000 users, and the bill keeps coming.
Loop company: every cohort of users generates 25% of its size in new users the next month. Month 1: 1,000 bought + 0. Month 2: 1,000 bought + 250 looped. Month 3: 1,000 + 312. The loop contribution keeps compounding; by month 12, total users materially exceed the funnel company's, and — crucially — if both companies cut spend to zero, the funnel company flatlines instantly while the loop company keeps growing for months.

The loop coefficient (here 25%) behaves like interest. Below 100% it amplifies acquisition; at 100%+ it goes viral. Most durable SaaS loops run between 15% and 40% — not viral, but a permanent discount on every acquisition dollar.

The four families of growth loops

Viral loops. Users expose the product to other users directly: invites, collaboration, shared artifacts. Fast cycle time, strongest in products with network utility.
Content loops. Usage generates indexable content — public profiles, templates, published pages — which ranks in search and recruits new users. Slow to start, extremely durable.
Paid loops. Revenue from one cohort funds acquisition of the next. Requires fast payback; this is the loop behind every efficient performance-marketing machine.
Sales loops. Customer success stories, case studies and champions who change jobs seed new accounts. Long cycle time, high value per cycle.

Most strong companies stack two or three loops with different cycle times: a fast viral loop for momentum, a content loop for durability, a paid loop for control.

How to design your first loop

Step 1 — Find the natural exhaust. What does product usage already produce that a non-user could encounter? Documents, links, exports, badges, emails, calendar events? The best loops amplify existing behavior rather than inventing new behavior.

Step 2 — Make the exhaust visible and attributable. A shared export with no branding recruits nobody. A shared page with a tasteful "made with" credit and a clear next step recruits.

Step 3 — Shorten the cycle. Loop growth rate depends on the coefficient and the cycle time. A loop that turns in 3 days compounds roughly 10 times faster than one that turns in 30.

Step 4 — Measure the loop as a loop. Instrument each edge: what fraction of users produce output, how many non-users each output reaches, what fraction of those convert. The product of the edges is your loop coefficient.

Step 5 — Simulate before you build. Before engineering anything, model the loop: if 30% of users share, each share reaches 3 people and 5% convert, your coefficient is 0.045 — too weak to matter. The model tells you which edge to strengthen first.

A worked example: turning a funnel motion into a loop

Suppose you run a reporting tool acquired mostly through paid search — a pure funnel. The loop-design questions, applied:

Exhaust: users generate weekly reports. Today those reports are PDFs emailed to stakeholders — invisible exhaust.
Redesign: replace the PDF with a live, hosted report link. Each report now has a URL a non-user opens in a browser, with a discreet product credit and a "create your own report" action.
Edges, estimated: 40% of active users share at least one report per month; each shared report is opened by an average of 4 stakeholders; 3% of viewers sign up. Loop coefficient ≈ 0.40 × 4 × 0.03 = 0.048 per month — small, but free, permanent, and improvable on three independent edges.
First improvement target: viewer-to-signup conversion (3%) is the weakest edge with the cheapest fixes — a contextual landing state and a one-click trial could plausibly double it, doubling the whole loop.

This is the general method: find the exhaust, put it on a URL, estimate the edges, attack the weakest one.

Common loop-building mistakes

Bolting a loop onto a product with weak retention. Loops amplify whatever exists. If users churn in a week, the loop amplifies churn-bound users.
Optimizing the coefficient and ignoring cycle time. Halving cycle time often beats a 20% coefficient lift.
Making sharing feel like spam. Loops that tax the user's social capital get switched off by the user. The output must be valuable to the recipient.
Measuring loops with funnel dashboards. If your analytics only show linear stage conversion, you will literally never see the loop.

Funnels and loops, together

The mature view is not "funnels are dead." Funnels remain the right microscope for any two-step conversion. Loops are the right telescope for the system. Use funnel math inside each loop edge; use loop thinking to decide what to build.

Mapping your loops explicitly — drawing the nodes, estimating each edge, simulating the compounding before committing a quarter of engineering — is the single highest-leverage planning exercise a growth team can run. Growth Pilot's visual loop builder and Monte-Carlo simulator exist for exactly this: sketch the loop, plug in your assumptions, and see whether it compounds before you build it.